Valdai Club

The desire to win the AI ​​race on the global stage as well as lobbying by tech investors and developers will push legislators to create conditions favourable to the development of innovation, while maintaining control over the big players. Antitrust legislation on AI in both the US and other countries should be monitored, as it will affect the global balance of power in the tech race, writes Anna Sytnik. The author is a participant of the Valdai - New Generation project. 

SInce the mid-19th century, Great Britain was the first country in the world to actively develop the automobile industry, which frightened the British public and led to the adoption of the "Red Flag Law". A specially trained person had to walk in front of the car on the road, waving a red flag and warning of danger. Quite naturally, this led to the fact that another country, in this case Germany, quickly pulled ahead in the auto market, showing a more progressive approach to the emergence of the new technology. 

Today, a similar situation is happening with the rapid development of artificial intelligence (AI). However, the present global competition is much more serious. If a car engine is just a technology that affects our mode of travel, then AI is an all-pervasive technology that changes life entirely. Therefore, lagging behind in the new technological race may prove critical in terms of further world political processes, which raises the stakes for both tech giants and the political establishment.

There are debates in global and national structures about the need to regulate the activities of technology companies. The problem is that in the current race in the field of large language models, excessive regulation of the activities of companies engaged in advanced developments may lead to the country lagging behind technologically on the world stage. At the same time, ill-considered steps to implement AI could put the state in an equally vulnerable situation both in terms of international security and in terms of monopolisation of the domestic market by the largest tech giants. This dilemma is reflected in the confrontation between two approaches - effective altruism and effective accelerationism in AI. Similar to the British Red Flag Act of 1865, a country that overregulates advanced development risks handing over the “palm of victory” to its strategic adversary. Only, while in the 19th century Britain lost out to Germany in the auto market, in the 21st century the “worst nightmare” of the United States, the country where the largest AI companies are located, is the possible leadership of the People’s Republic of China. In this regard, the ongoing process of developing a regulatory mechanism for AI in the United States is particularly interesting, since it can both weaken the position of American technology companies in world politics (which still have enormous influence) and become a model for the application of similar restrictive measures in other countries that are oriented towards the United States.

The Problem with the Antitrust Regulation of AI

AI is an end-to-end technology where data, computing power, and algorithms play a major role. From the point of view of regulators, today's AI systems are mostly "black boxes". Not only do not all legislators understand how complex large models work, not all developers understand them. In addition to their closed nature, AI differs from other, more traditional areas in the speed of development and network effect. Legislation does not have time to adapt, which is effectively used by technology companies, and the rapid increase in the number of users of AI services increases the dominance of large players, in whose hands the main digital resources are already concentrated. The lack of adequate regulation of AI can have negative socio-economic and political consequences: the gap between tech giants and the rest of the economy will widen in the world, and their global political influence will only increase. These challenges require the development of new methods of antitrust regulation that take into account the specifics of large models, as well as international cooperation to create common standards and rules. 

However, it is too early to talk about universal rules for the antitrust regulation of AI. Major international organisations are still busy discussing topics such as AI ethics, potential risks to security and human rights. Strategies for the antitrust regulation of AI are developed at the national and regional levels and vary depending on the legal framework. The European Union, in its AI Act, has put forward the strictest requirements for transparency and accountability for AI developers in order to prevent the creation of monopolies. China controls its AI ​​sector, focusing on regulating the abuse of data collection and its use for training AI. Other BRICS countries, among which Russia takes the most active position, are developing a legal framework for regulating AI, seeking to pay attention to security issues without compromising the development of innovation. The United States, on the other hand, focuses on practices of abuse of position in the AI ​​market, but its desire to maintain global leadership in the technology race prevents the adoption of adequate antitrust legislation.

Effective altruism vs. effective accelerationism in AI

There are two philosophical approaches that help to understand the essence of the current contradictions in approaches to AI regulation – effective altruism and effective accelerationism. It is important to note that they are not completely opposed and in certain cases complement each other. In simpler terms, supporters of effective altruism focus on long-term consequences and ethical issues, while adherents of effective accelerationism strive for rapid implementation of innovations, believing that technological progress itself will bring significant benefits to society. An example of effective altruism is the AI ​​safety movement, which culminated in a joint statement by “tech leaders” led by Elon Musk about the need to stop the development of AI in March 2023. However, a year later the AI ​​Safety movement was declared “dead”. 

There are two premises behind such a loud statement. The first is the new federal policy guidance document “Driving US innovation in Artificial Intelligence” guaranteeing broad federal support for AI research and consistent recognition of the importance of maintaining US leadership in AI for national security. The contents of this document allow us to conclude that lawmakers have decided that it is better to accept the risks associated with American AI systems than to leave Chinese developments without an American counterweight. The second prerequisite for the possible “victory” of effective accelerationism is a policy change within the largest technology companies towards secrecy and the active release of new, increasingly powerful versions of their models, i.e. an intensification of the race.

However, not everyone agrees with the strategy of effective accelerationism. A “red flag” for the development of AI in the United States could be the tough stance US Federal Trade Commission (FTC) Chairperson Lina Khan has on antitrust law against the modern tech giants. Her figure deserves attention for one simple reason - her fight against the “mafia bosses” in the large tech AI industry in the United States can have an impact on the global tech landscape. Thus, in January 2024, the FTC launched an investigation into corporate partnerships and investments with the suppliers of five US companies engaged in generative AI - Alphabet, Inc., Amazon.com, Inc., Anthropic PBC, Microsoft Corp. and OpenAI Inc. These corporations control the resources that startups and other companies around the world rely on to develop and deploy AI tools, including cloud services and computing power, as well as vast amounts of data. In addition, if the tech giants maintain their monopolistic control, in which the business model is based on a policy of minimum prices and does not allow new “big ideas” to be born, the United States may cease to be the “homeland” of the world's leading technologies. Of course, this revealing position has received a lot of criticism. Tech investors and developers point out that excessive application of antitrust laws is a bad economic policy. As an argument, they cite the example of the European Union, which has regulated its economy so much that it has led to a lag in technological areas. The second argument is a reference to the techno-economic war with China for the best technologies for the next 20 years (and, of course, the need to win it). Finally, the third reason that, in their opinion, AI regulation should be relaxed, is that the rapid achievement of the best possible AI technologies will lead to great abundance, significant growth in GDP, and increased productivity.

So, although the voices of effective altruism are heard and come from very influential figures, they are consistently drowned out by the cries of supporters of effective accelerationism. The standoff continues.

***
A new round of technological progress is already influencing social processes, which poses a difficult task for legislators to balance between innovation regulation and stimulation. In the United States, this situation is extremely problematic, since American tech giants enjoy great power both in the country and throughout much of the world. They do not want to lose it. However, the current stage of AI implementation requires greater attention to the safety of AI models, which means that the world is on the verge of new changes. The desire to win the AI ​​race on the global stage as well as lobbying by tech investors and developers will push legislators to create conditions favourable to the development of innovation, while maintaining control over the big players. Antitrust legislation on AI in both the US and other countries should be monitored, as it will affect the global balance of power in the tech race.